runPredict¶

PredictResource.runPredict(request, sub_analysis_id, **kwargs)¶

Prior mandatory steps 1) Upload dataset 2) Create analysis 3) Create sub analysis 4) DataSharp 5)iLearn

This function performs the predictions. When a new data (eval data) is uploaded that has same number of attributes as dataset used for the analysis, this function performs the predictions on this new dataset and adds an additional column with predicted value for each record. There can be exceptions where a record may not have prediction because Record may have a new level that is not seen in training data. The returned data with predictions marks such records as could not predict.

User can pick any specific model for predictions using model number provided for each model out of iLearn function. This function is a background function that generated predictions and returns a prediction_id. Use getPredict function after this to obtain the actual prediction output.

Arguments

sub_analysis_id	Give sub analysis id
model_no	Give model number from model list
data_file	Give path of eval dataset file

Possible errors

Error message
Invalid sub analysis id
Please wait for a while. Evaluation dataset prediction is still running
Please wait for a while. Predict is still running
Please run ilearn first

POST Request Example

curl -u username:password -X POST -F "model_no=1" -F "data_file=@/path/to/eval.csv" {url_prefix}/predictions/{sub_analysis_id}/

Response Example

{
    "error": false,
    "error_msg": "",
    "result": {
        "prediction_id": 463,
        "prediction_accuracy": 91.66,
        "prediction_data":[
            {
                "TargetdataCleaned": "no",
                "age": 36.0,
                "balance": 2843.0,
                "campaign": 1.0,
                "contact": "cellular",
                "day": 12.0,
                "default": "no",
                "duration": 473.0,
                "education": "secondary",
                "housing": "no",
                "isCleaned": "Yes",
                "job": "blue_collar",
                "loan": "no",
                "marital": "divorced",
                "month": "feb",
                "pdays": 182.0,
                "poutcome": "success",
                "predictedClass": "yes",
                "previous": 1.0,
                "y": "no"
            },
            ....
        ]
    }
}

Error Response Example

{
    "error": true,
    "error_msg": "Please wait for a while. Evaluation dataset prediction is still running",
    "result": {
        "prediction_id": 463,
    }
}